zopfli-rs

Bindings to zopfli for deflate, gzip, and zlib compression.

Purpose

zopfli is already implemented in pure Rust by @carols10cents (carols10cents/zopfli), so it might seem duplicative to create bindings to the original C / C++ library by Google (google/zopfli). In order to give better insights into the performance of the pure Rust library and to give a slightly faster alternative, zopfli-rs was made.

Benchmarks

zopfli-rs just slightly inches ahead of zopfli when it comes to gzip compression. Both zopfli-rs and zopfli follow a strange performance curve that seem to be faster for certain sizes of data (in this case ~10KB is the sweet spot). Changing up the Options would more than likely improve these results, but that would take a bit of time to tweak to perfection.

test bench_zopfli_1         ... bench:   5,026,521 ns/iter (+/- 547,023)
test bench_zopfli_10        ... bench:   5,817,841 ns/iter (+/- 585,086)
test bench_zopfli_100       ... bench:  19,293,553 ns/iter (+/- 2,106,929)
test bench_zopfli_1000      ... bench: 290,497,031 ns/iter (+/- 32,570,971)
test bench_zopfli_10000     ... bench:  38,902,158 ns/iter (+/- 6,040,615)
test bench_zopfli_100000    ... bench: 301,934,186 ns/iter (+/- 17,868,991)
test bench_zopfli_rs_1      ... bench:   2,439,766 ns/iter (+/- 644,219)
test bench_zopfli_rs_10     ... bench:   3,299,515 ns/iter (+/- 359,117)
test bench_zopfli_rs_100    ... bench:  14,146,432 ns/iter (+/- 1,505,380)
test bench_zopfli_rs_1000   ... bench: 209,671,685 ns/iter (+/- 9,135,909)
test bench_zopfli_rs_10000  ... bench:  29,229,623 ns/iter (+/- 3,060,932)
test bench_zopfli_rs_100000 ... bench: 289,308,248 ns/iter (+/- 13,868,834)